Highlighting areas of improvement. Based on the steps in the funnel, we documented issues from customer support tickets and used FullStory to gain quantitative and qualitative insights. Once we pinpointed these problems, we began proposing high-impact solutions that required minimal engineering effort.
Guiding hosts to the correct signup type. Kopa’s architecture requires two account types: hosts and renters, each with separate functions. Hosts accidentally signed up as a “renter” instead of a “landlord,” making them unable to move forward and prompting them to contact customer support. We made the choice more clear by structuring each option as a sentence and adding illustrations.
Reducing steps and complexity in onboarding. Because of the high number of scams in the rental industry, we needed a government-issued ID and selfie verification from each host. Because Kopa is a newer company, hosts didn’t trust us yet and were wary to provide that sensitive information. We saw a 40% dropoff at this step. We removed this step and only required it to publish a listing.
Secondarily, we saw hosts dropping off throughout various stages, which told us onboarding was too long. If we didn’t need the information up front to create their account, we made the step skippable.
Defining and analyzing steps in the funnel. We define our host funnel from creating an account to publishing a listing. We used a combination of FullStory and Google Analytics to determine the flow dropoffs at each step.
Revamping onboarding to make it more approachable. With 70%+ of our hosts identifying as age 50+ and female, we structured the flow as questions and added a photo of a friendly female face to make the host feel like they were being handheld through the onboarding process. At the end of onboarding, we reduced options for the CTA and only showed one, “Create a listing,” so hosts knew exactly what to do next.
Small UI fixes to make progress clear. Adding a listing requires a lot of information. We broke up the process into 8 steps, but we noticed 12% of hosts dropped off after step 1, not realizing they had to complete more steps. We added numbers to steps to the navigation bar and to each button and changed “Save and exit” to a link to deprioritize that action.
Significantly reducing the time hosts spend on listings. The 8 required steps took hosts an average of 14 minutes to complete. Since we needed all of that information and because many of the hosts on Kopa listed on Airbnb, we built an Airbnb importer. While this feature was a larger engineering feat, the time to add a listing decreased to 4 minutes and Kopa gained thousands of listings through the importer.
The biggest challenge of this feature was that the importer took about 1 minute to import. For hosts with dozens of listings, we added several messages along the way to let them know if we were still importing and sent an email notification when the import completed.
We also made labels for Drafts stand out more in the UI to catch the user's attention.
Asking for information at the timeliest moment. We put the ask for all sensitive information in the final step of creating a listing. Since hosts had spent at least several minutes going through this process, they were most likely to give us the required information in this moment.
Tracking success. Each of these improvements was treated as its own project and watched the metrics for each of these changes. We increased conversion by 17%, which was a huge win. The most impactful changes were removing the ID verification from onboarding and adding the Airbnb listing import functionality.
Summary
The biggest challenge of this project was determining the problems that needed to be solved. Without a world-class data analysis setup, we had to be scrappy, manually looking through customer support tickets and FullStory sessions.
Before making these changes, our supply was focused in the San Francisco Bay Area. These improvements helped us expand to over 100 U.S. cities and gain thousands of hosts and listings.
